EndoMetronome

    EndoMetronome is a machine learning system which predicts timing, gene expression, and risk of pregnancy loss from luteal phase endometrial biopsies.

    This website provides instructions to the visualisation server which is available here.

    The website also allows the input of a heterogeneous graph to return predictions for all outputs. For an example graph, please download this file.

    For any queries, please contact Fayyaz.Minhas@warwick.ac.uk

    George Wright, Thomas M. Rawlings, Mark Eastwood, Paul Brighton, Mireia Taus Nebot, Anja Estermann, William T. M. Flett, Awais Younis, Komal Makwana, Hiroyuki Yoshihara, John D. Aplin, Chow-Seng Kong, Mark Christian, Emma S. Lucas, Joanne Muter, Jan J. Brosens and Fayyaz Minhas Whole slide image analysis of the endometrial decidual reaction reveals multiscale perturbations associated with miscarriage. bioRxiv.

Overview of EndoMetronome model architecture

How to Use The Visualisation Tool

  • Select the slide at the top-right of the window.
  • Use the "Add Overlay" button to choose output for viewing.
  • Switch heatmaps with the 'color by' selector.
  • Red indicates high predictions; blue indicates low (in default coolwarm color map)
  • Adjust overlay transparency with the 'overlay alpha' slider.
  • See example slides below.
Guide illustration for using the visualization tool

Example Heatmaps

Endometrial Timing Prediction

EndoTime
Low High
LH+
Low High

Stromal Gene Prediction

DIO2
Low High
SCARA5
Low High
IL15
Low High
PLA2G2A
Low High
TIMP3
Low High
IGFBP1
Low High
PLA2G2A/DIO2 Ratio
Low High

Glandular Gene Predictions

SLC15A2
Low High
CXCL14
Low High
GPX3
Low High
DPP4
Low High
GPX3/SLC15A2
Low High

uNK Cell Gene Predictions

IL2RB
Low High
ITGAD
Low High
TAGAP
Low High

Upload A Graph!